ChanServ changed the topic of #wayland to: https://wayland.freedesktop.org | Discussion about the Wayland protocol and its implementations, plus libinput
tzimmermann has quit [Read error: Connection reset by peer]
Brainium has joined #wayland
mtj has quit [Quit: mtj]
nerdopolis has quit [Ping timeout: 480 seconds]
tzimmermann has joined #wayland
iomari891 has joined #wayland
riteo has quit [Ping timeout: 480 seconds]
mwhoosier has joined #wayland
<mwhoosier>
pq: Thought maybe I could catch you here rather than mess around on wayland-devel for a moment.
<pq>
mwhoosier, very quickly at this time, I'm near my end of day. :-)
<mwhoosier>
pq: Haha, okay. Should be quick! Just wanted to close out this topic about the restrictions on making updates to the KMS scene that a writeback connector is fed by. One (implied?) suggestion from the dri-devel people was to switch e.g. the DP-1 connector to a different CRTC while the writeback connector drains out its previous frame. (a) Is that even legal to bounce back and forth between two CRTCs repeatedly? (b) Would that lead to a modeset on each f
<pq>
mwhoosier, your message was truncated for being too long. But yeah, re-routing connectors takes a modeset I believe, and you would still be touching the old CRTC because of taking the connector away from it.
<Calandracas>
does color-lcms work for all backends? If I run weston as a nested wayland client (for example, as a kiosk shell within sway), would the colour pipeline still apply the appropriate transformations?
<mwhoosier>
pq, Okay, thanks!
<pq>
mwhoosier, I read the discussion, and while I don't remember my interpretation of that comment, I did not interpret it like you did.
<mwhoosier>
pq: you mean Daniel Vetter's remark about using a different CRTC?
<pq>
yeah
<pq>
oh, I think it was about using another CRTC from the beginning, all the time
<pq>
If you had enough planes, you could program two different CRTCs the same way, for different connectors.
<mwhoosier>
I thought that my interpretation was probably wrong, so that's fine. I've never seen any DRM master code bounce around between CRTCs between frames, so it seems a longshot that's what Daniel was saying.
<pq>
mwhoosier, yeah, I think it needs a modeset, and that ruins everything for a start.
<mwhoosier>
Huh, I'm not immediately seeing why having two different connectors would help anyway. It's a specific physical output whose contents we would want to stream anyway.
novenary has quit [Remote host closed the connection]
<pq>
Calandracas, umm... IIRC, yes
<mwhoosier>
pq: Anyway, thanks for all the help on this one.
novenary has joined #wayland
<pq>
Calandracas, color-lcms is not specific for any backend itself. It's up to the frontend to configure output profiles and enable color management.
<pq>
Calandracas, color-lcms does depend on GL-renderer and GL ES 3.0.
<pq>
mwhoosier, I mean program two different CRTCs to produce the same image; one feeds the real connector, another feeds the writeback connector.
<pq>
then they can update independently
<pq>
writeback may still be low frame rate though
<mwhoosier>
pq: Oh, okay. Fair. But that won't help with the secret use-case I talked about in the email, where some other virtualized OS is stuffing contents into the planes behind KMS's back.
<pq>
indeed
<pq>
dri-devel people didn't get that memo yet, I think :-)
<daniels>
shhhh
<mwhoosier>
No need to horrify them...
<mwhoosier>
It's not as though I don't realize what a ghastly corporate engineering thing this is ;)
<Calandracas>
pq, wow this is awesome. For some reason I didn't think of running applications like gimp or rawtherapee within weston as a nested wayland client, but this looks like it works very well
iomari891 has quit [Ping timeout: 480 seconds]
<pq>
Calandracas, heh, as long as you don't want to show bigger than sRGB gamut. :-)
<pq>
Calandracas, OTOH, if you can set up color management in the app, x11 style, that should be usable too, without Weston in between. The only assumption needed is that the desktop does not mangle pixels, or at least does not change how it mangles pixels.
<pq>
The X11 style color management has always been possible in a Wayland-based DE, maybe sans VCGT.
iomari891 has joined #wayland
<pq>
When you use Weston's color management, it assumes all windows are sRGB when protocol says nothing else. And there is no app using the protocol yet, aside from weston-image in git main branch.
<pq>
(Weston does not implement the parametric image description interfaces yet.)
<Calandracas>
hum
<Calandracas>
yeah I have a DCI-P3 monitor so would want to full gammut
<pq>
then you need apps to tell Weston, when they are targeting full gamut - protocol is needed for that, in the Wayland color management style
KarenTheDorf has quit [Ping timeout: 480 seconds]
<pq>
in X11 style, you can make it just work for apps that do color management themselves, and all other apps will look too saturated.
KarenTheDorf has joined #wayland
<pq>
When testing Weston from git, if you have PNG or JPEG with embedded ICC profiles, showing them with weston-image while setting a P3 ICC profile for the output in Weston, should make a nice demo.
<pq>
unfortunately, that is the extent of the implementation we have in Weston so far
bnason2 has quit [Read error: Connection reset by peer]
bnason2 has joined #wayland
melonai has quit []
systwi has quit []
systwi has joined #wayland
iomari891 has quit [Ping timeout: 480 seconds]
iomari891 has joined #wayland
Calandracas_ has quit [Ping timeout: 480 seconds]
melonai has joined #wayland
Calandracas has joined #wayland
sima has quit [Ping timeout: 480 seconds]
feaneron has joined #wayland
rv1sr has quit []
coldfeet has quit [Remote host closed the connection]
soreau has quit [Ping timeout: 480 seconds]
soreau has joined #wayland
Brainium has joined #wayland
mvlad has quit [Quit: Leaving]
karolherbst has quit [Ping timeout: 480 seconds]
karolherbst has joined #wayland
iomari891 has quit [Read error: Connection reset by peer]
julio7359 has joined #wayland
feaneron has quit [Ping timeout: 480 seconds]
nerdopolis has joined #wayland
CME_ has joined #wayland
CME has quit [Ping timeout: 480 seconds]
chiku has joined #wayland
Sid127 has quit [Read error: Connection reset by peer]
<KarenTheDorf>
So... there's the cursor shape protocol in staging. But there's no point in it if you have to support non-supporting compositors and implement a fallback to wayland-cursor or something? Am I missing anything?
<Nefsen402>
The benefit of cursor shape is to have consistent size/style of the cursor
<Nefsen402>
Compositors and clients may handle scaling differently and so the cursor may change sizes as the cursor changes from a compositor controlled one to a client one
<Nefsen402>
Also, basically no client supported animated cursors
<Nefsen402>
With cursor shape, as long as the compositor supports it, it will work
<KarenTheDorf>
So I should probabally change to cursor shape where supported?
<KarenTheDorf>
I go by XCURSOR_SIZE in the environment for the size? Is that not always correct? (I suppose it can't be, on multi-monitor setups with per-monitor scaling)
<Nefsen402>
XCURSOR_SIZE afaik isn't even a stardard, it just sort of a thing clients started doing
<Nefsen402>
and when you throw display scaling (hidpi) things get more complex
chiku has quit [Ping timeout: 480 seconds]
<Nefsen402>
You should support the protocol if possible
<alice>
i started using 1.5 scale and now i see like 3 separate cursor sizes... the only thing that gets it right is qt and things that support cursor-shape
<alice>
definitely support if if you don't mind maintaining two paths :P
<KarenTheDorf>
Well, wayland is copying the most popular OS there with having 3 different sizes for everything depending how well applications support hidpi scaling.
<Nefsen402>
it isn't a big deal at all
<KarenTheDorf>
Supporting shape looks a lot simpler than wayland-cursor, it's just not ubiquitous support. I assume Mutter is not going to start supporting it since it's not a problem for GTK based apps.
<alice>
there's an open pr for mutter actually
<alice>
they probably will
<alice>
and if mutter does gtk will probably support it as a client too
<Nefsen402>
ofc mutter doesn't support it yet
<Nefsen402>
Why am I surprised
<KarenTheDorf>
Over on the vulkan discord, gnome users asking "How do I get (non-awful) window decorations on wayland" is a perrenial question.
nerdopolis has quit [Ping timeout: 480 seconds]
<KarenTheDorf>
Normally point them at libdecor now, or suggest just running on fullscreen if they are on GNOME and don't/can't use XWayland.
<Nefsen402>
There's a solution for that: libdecor
<Nefsen402>
But it's a client side thing
<KarenTheDorf>
Indeed, I just said that :)
<Nefsen402>
Indeed
CME_ has quit [Ping timeout: 480 seconds]
<KarenTheDorf>
100% get why CSDs should be the default though, they are a lot nicer from a compositor's POV. I'm just not a compositor author.
<Nefsen402>
There are a couple of advantages to purely SSDs like for tiling window managers